The questions that stall a people counting rollout are rarely the ones covered in vendor brochures. They surface in week three of a pilot, when the operations director asks why Tuesday's count in the flagship store is 11% higher than the POS-derived estimate, or when procurement wants to know what happens to the data if the sensor manufacturer is swapped out in year four. This people counting FAQ addresses the questions we actually hear from buyers, support teams, and end users once contracts are on the table — not the introductory ones.
Be suspicious of any vendor guaranteeing a flat 99%. Accuracy depends on conditions the vendor does not fully control: lighting, entrance layout, ceiling height, and how visitors actually move. Groups clustering in a wide doorway behave differently from single-file traffic through a narrow one. The honest framing is a contractual minimum of 96%, with 98–99% typically achieved when conditions allow. That distinction matters in negotiation: the minimum is what you can enforce; the typical range is what you should expect at most doors after proper commissioning. Insist that the contract also defines how accuracy is validated — manual count audits over defined time windows, not a one-off spot check on installation day.
Yes — unless you exclude it deliberately. In a mid-size store, employees crossing the entrance for deliveries, breaks, and trolley returns can inflate footfall by 5–15%, which quietly deflates your conversion rate and misleads staffing models. Staff-exclusion algorithms identify and remove employee movements from the count, and this should be switched on and validated per site, not assumed. A practitioner's tip: the stores where staff exclusion matters most are the ones where the stockroom door sits near the customer entrance. Walk every entrance during a site survey and note staff paths before the sensors go in — retrofitting exclusion zones after a month of polluted data means a month of baselines you cannot trust.
The better question is whether you should standardise at all. Enterprise estates are messy: an acquired chain arrives with Hikvision cameras already on the ceiling, a new flagship justifies premium 3D units like Xovis, and smaller kiosk locations only warrant a Milesight or AXIS device. A sensor-agnostic platform lets you keep mixed hardware feeding one reporting layer, which protects you in two ways: you avoid ripping out functioning devices during migration, and you retain pricing power when hardware contracts renew. Locking analytics and hardware to a single manufacturer is the most common regret we hear from buyers three years into a deployment.
Modern AI sensors can estimate age and gender profiles and separate children from adults in the count. Two practical notes. First, excluding children changes your conversion rate materially in family-heavy formats — a toy retailer counting a family of five as five potential buyers is measuring the wrong denominator. Second, demographic estimation should be handled as anonymised, aggregated data; involve your DPO early so privacy review does not become the bottleneck at the end of the project rather than a checkbox at the start.
Both models are workable; the deciding factors are usually your IT security policy and where the platform must integrate. If footfall feeds ERP-based labour planning or a BI environment like Power BI, confirm the integration path before signing — API availability, export formats, and refresh frequency. On ownership: the counts your stores generate are your operational data. Contracts should state explicitly that historical data is exportable in full, at no additional cost, at any point including termination. If a vendor hesitates on that clause, treat it as a signal.
Enterprise buyers increasingly ask what comes after door counting. In-store movement tracking (VemTrack), zone and space utilisation (VemSpace), and tenant-level analytics for shopping centres (VemTenant) all build on the same sensor infrastructure. The sequencing advice from implementers is consistent: get entrance counting stable and trusted first. A team that doubts the door counts will not trust heatmaps built on the same devices. Six months of clean, validated footfall data is the credibility foundation everything else stands on.
Three things, in our experience. A proper site survey per location rather than a one-spec-fits-all install plan. A named data owner on the retailer's side — someone accountable for checking data quality weekly during the first quarter. And a validation protocol agreed before installation, so that "is it accurate?" has a defined answer rather than an argument. Vendors with long track records — Vemco has been building people counting and retail analytics software since 2005 — will have these processes documented; ask to see them before you commit, because the documentation quality tells you what support will feel like in year two.
Have a question this FAQ did not cover — about accuracy validation, mixed sensor estates, migration from a legacy counting system, or contract terms worth negotiating? Contact the Vemco Group team and get a direct answer from people who have handled these exact questions across enterprise deployments.